Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
nuz
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
8 ms
·
31.
▲
by
nuz
2y ago
What are you using it for? Curious if there's any interesting purposes I haven't thought of
32.
▲
by
nuz
2y ago
OpenAI are so good at fake leaking things to build hype
33.
▲
by
nuz
2y ago
I wonder how overfit it is though. You could fit a lot of doom resolution jpeg frames into 4gb (the size of SD1.4)
34.
▲
by
nuz
2y ago
Looks like it's already suffered from 'issue bloat' of people suggesting small changes in the issues and the maintainer not being tough enough to reject the changes
35.
▲
by
nuz
2y ago
Colab arguably
36.
▲
by
nuz
2y ago
I can never read comments on any wolfram blog on HN because they're always so mean spirited. I'm seeing a nerdy guy explaining things from a cool new perspective I'm excited to read through. The comments almost always have so
37.
▲
by
nuz
2y ago
Americans are assumed to not make use of much of any of that unlimited vacation, whereas in europe it's actually ok to
38.
▲
by
nuz
2y ago
I know X is majority hated these days on here but this sounds like a great victory for all if they win imo
39.
▲
by
nuz
2y ago
They've cut out things like viewing who liked a tweet etc (for cost savings I suspect) so the site is probably dirt cheap. Can't imagine ads don't pay for the cost of running servers at this point, and a headcount reduction m
40.
▲
by
nuz
2y ago
~3 years since gpt-4 was finished training in 2021
41.
▲
by
nuz
2y ago
Investors betting on openai wrappers with no moat outraged at losing money.
42.
▲
by
nuz
2y ago
The weird logic of elon being controversial so this somehow is a good thing
43.
▲
by
nuz
2y ago
You can make this without loras too (e.g. elon and AOC and many celebrities exist without extra steps). Wonder how this affects open source models in general
44.
▲
by
nuz
2y ago
It'll move to lawless eastern europe wild west probably.
45.
▲
by
nuz
2y ago
Lots and lots of synthetic data from the bigger models training the smaller ones would be my guess.
46.
▲
by
nuz
2y ago
Everyone complaining about not having data access: Remember that without meta you would have openai and anthropic and that's it. I'm really thankful they're releasing this, and the reason they can't release the data is o
47.
▲
by
nuz
2y ago
I like lower case tweets and texts but lower case in articles like this is just ridiculous (and trying too hard to be cool)
48.
▲
by
nuz
2y ago
I'd like to continue on the narrative here but the truth is kind of that tech itself hasn't really had much exciting things to report on in the ~2017 to now period. If I were in their shoes I'd probably switch reporting to an
49.
▲
by
nuz
2y ago
That's not mastodon fyi
50.
▲
by
nuz
2y ago
Almost rorschach painting levels of ambiguity in that thing
51.
▲
by
nuz
2y ago
Was hoping this was using the built in gemini web api discussed on HN a few days ago (I guess it's not multimodal though).
52.
▲
by
nuz
2y ago
This is tiktoks scraper? How come tiktok does mass scraping of websites?
53.
▲
by
nuz
2y ago
Just switch to duckduckgo. Silly name but it's so worth it.
54.
▲
by
nuz
2y ago
https://github.com/wjakob/instant-meshes
55.
▲
by
nuz
2y ago
Such a silly argument. Fixing topology is a nearly solved problem in geometry processing. (Or just start with a good topology and 'paste' a texture onto it like they develop techniques for here.)
56.
▲
by
nuz
2y ago
Use SDXL instead of sd1.5. Also don't do negative prompting on "distorted, ugly" because everyone follows these guides blindly and the result is the same types of mode collapsed faces which people have learned to recognize as
57.
▲
by
nuz
2y ago
They're pretty clear about being pro safety to the extreme, and mass surveillance to protect american interests and abuse of LLM tech (e.g. open source misuses) are probably within the umbrella of ends justifying the means logic anthro
58.
▲
by
nuz
2y ago
They probably still keep the data to sell to LLM training organizations
59.
▲
by
nuz
2y ago
I kind of respect deepmind for simply keeping their nose to the ground and doing good work like this without overhyping it too much. More of the under promise over deliver engineering style than some of their competitors tends to do it
60.
▲
by
nuz
2y ago
Miss when engadget was pro-tech
More ›